AI Workflows FAQ

What is AI Workflows?

AI Workflows is a "fair-code" workflow automation tool that allows University of Michigan faculty and staff to automate repetitive tasks and connect academic tools.

What data can I pass to AI Workflows?

AI Workflows is suitable for institutional data in the Low and Moderate sensitivity categories only. 

For example, FERPA data is classified in the Moderate category and may be passed to n8n workflows. HIPPA data, however, is in the High sensitivity category and may not be used in AI Workflows. 

Who can use AI Workflows?

All faculty and staff have access to AI Workflows. To gain initial access to the service, all users must complete the AI Workflows Training Canvas course.

How does AI Workflows use AI?

AI Workflows can use LLMs available via the U-M GPT Toolkit. Please note that LLM usage is billed separately from any costs you incur running AI Workflows.

Do failed scheduled workflow executions count against my usage limit?
Yes. Every time a scheduled workflow attempts to run, whether it succeeds or fails, it counts as an execution. It is the responsibility of the workflow owner to monitor their "Execution List" and ensure workflows are in good working condition.
 
Note that manually triggered workflow executions never count against you usage limit.

What happens if my workflow keeps failing? To maintain platform stability, ITS reserves the right to intervene if a workflow fails repeatedly (typically after 10 consecutive failures).
 
  • For Scheduled Workflows: If a workflow triggered by a schedule (e.g., every hour) continues to fail, ITS reserves the right to disable the Schedule Trigger. * Resolution: If your trigger is disabled, you must fix the underlying error and manually re-enable the trigger to resume the automation.
Can students use AI Workflows?

AI Workflows is not open for student usage at this time. Only faculty and staff may use AI Workflows.
 

Do I need to know how to code?

No. Most of AI Workflows is "drag-and-drop." You use a visual editor to connect boxes (called "nodes") with lines. However, if you want to add custom logic, AI Workflows lets you add code to specific nodes.

Are there any example templates?

Yes.  AI Workflows includes a U‑M Templates project with starter “skeleton” workflows you can duplicate to jump-start your own. All users get access to it within 24 hours of signing in for the first time (Projects → U‑M Templates).

How do I obtain credentials for various services?

Refer to our Knowledge Base articles on our Support page for more information